Games

New research tool predicts what online gamers will do next

A team of researchers, utilizing data gathered from the behavior of 14,000 online gamers, were able to accurately predict what actions gamers would take in specific situations (Photo: R. Pollard)
A team of researchers, utilizing data gathered from the behavior of 14,000 online gamers, were able to accurately predict what actions gamers would take in specific situations (Photo: R. Pollard)

Is there such a thing as free will, or are our actions predetermined by the way our brains work? If recent research conducted at North Carolina State University is anything to go by, it might seem that the latter is more likely to be true - at least when it comes to gaming. After analyzing the behavior of 14,000 players of the online role-playing game World of Warcraft, an NCSU team was able to predict the future actions of those players with up to 80 percent accuracy.

In World of Warcraft, players earn virtual badges for achieving various goals. The researchers analyzed data on which badges individual players had earned, and in what order. Patterns began to emerge, in which certain achievements were shown to have a recurring correlation with other specific achievements - if a player did A, then X number of steps down the road, they were likely to do B.

These related achievements tended to bunch into groups known as cliques. Any player that had attained some of the achievements in one clique were found to be likely to pursue other achievements in that same clique. Any one clique could consist of over 80 different achievements.

Some of that predictability, of course, could simply be due to logistics - once players achieve a certain status via specific actions, they are then able to perform certain new actions, so they proceed to do so. Some of the linked achievements within cliques, however, appeared to have no obvious logical connection - earning a badge for skill in unarmed combat, for instance, was shown to have a high correlation with earning a badge for world travel.

"A good game stands on its own," said Dr. David L. Roberts, an assistant professor of computer science at NCSU. "If you want to improve it, you have to make sure players will like any changes you make. This research can help researchers get it right, because if you have a good idea of what players like, you can make informed decisions about the kind of storylines and mechanics those players would like in the future."

"This work could obviously be used for World of Warcraft or other MMORPGs [massively multiplayer online role-playing games], but it also applies to any setting where users are making a series of decisions. That could be other gaming formats, or even online retailing."

The study raises at least one interesting question: is our behavior in the real world equally predictable?

  • Facebook
  • Twitter
  • Flipboard
  • LinkedIn
2 comments
teeduke
Are you Poker players and Casino operators paying attention? Or do we have to spell it out for you?
Bert Johnson
This seems terribly sloppy work. If one researches the achievement trees of any particular game and evaluates the realtionships, it\'s not the magic it sounds. The two \'unrelated\' achievements quoted here (one which was removed last year) are in fact, quite related: both require maximum game level, and neither are considered essential to gameplay (there is no instance where unarmed combat is used in the game, nor every nook and cranny of the game map has relevvant content). Therefore, they would fall into a handful of achievements with the same requirements and interests. This is about as magical as a study showing people that buy conditioner tend to buy shampoo. The majority of all achievements are in fact interrelated. such as reaching certain levels or going to certain regions/instances that are also level dependent. The \'dungeon\' achievements are all sequential, and going to any WoW gaming site would show exactly which ones come directly before and after other related achievements (character level, level dependant achieves such as \'dual talent\' and \'flight\'). Since there are overwhelming numbers of \'alts\' (alternate characters for one player) that are started and abandoned, the standard model of progression and related achievements would dominate the data, leaving any actual \'choices\' exclusively in the 20% of unpredictability range. As certain individual collection achievements would be more likely to appear right before an over-arching collector titles, and profession related achievements would also tend to show up in sequential orders as people \"power level\" their alts, one begins to see that a huge portion of this data is simply sequential and has no bearing on choices made. To pass off naturally sequential activities is akin to \'predicting\' someone will brush their teeth because of observed behavior that happens to be them applying toothpaste to a toothbrush. Neither is a meditation on free will, just sloppy science.